## _wp1623

## Source details

**Canonical URL:** [_wp1623](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp1623.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp1623.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp1623.pdf.json)

---

### Introduction and motivation
- Imported inputs identified as the dominant transmission channel of import-price shocks to domestic prices because:
  - Consumption goods constitute at most a little over 10 percent of total imports in Korea (Figure 1).
  - Imported goods cover around 7% of Korea’s CPI basket (informal KOSTAT interview).
  - Correlation coefficient of 0.97 between monthly change of the aggregate import price index and that of the aggregate imported input price index in Korea (Figure 2).
  - Correlation of 0.63 between monthly change of the aggregate producer price index and that of the aggregate import price index; import price index is about four times more volatile than the producer price index (Figures 4–5).
- Paper focus:
  - Estimate imported input cost pass-through to domestic producer prices using sector-level monthly data and an error correction model (ECM).
  - Explicitly construct sector-level imported input price indices using bilateral sector-level Input-Output (IO) tables and sector import prices.
- Distinct contributions:
  - Directly estimate pass-through from imported input prices to sector-level producer prices.
  - Separate roles of imported input cost shares and markup adjustments in determining effective pass-through.

### Econometric framework (baseline model and identification)
- Pricing and cost structure:
  - Firm price set as markup over marginal cost.
  - Marginal cost modeled as a Cobb-Douglas aggregator of domestic intermediate input price (P_dom), imported intermediate input price (P_imp), and unit labor cost (ULC) with shares α, β, and γ.
  - Effective imported input cost pass-through to producer prices = (1 + ζ_β)·β where β is imported input share and ζ_β captures markup adjustment in response to imported input cost shocks.
- Empirical construction of effective input prices:
  - Effective imported input price EIMP_it = weighted average across input sectors of sector import price indices using imported-input shares β_ij from IO tables.
  - Effective domestic input price EDOM_it = weighted average across input sectors of sector domestic-input price indices using domestic-input shares α_ij.
  - Effective labor cost term constructed as γ_i · ULC_it.
  - Baseline EDOM omits own-sector domestic input price to reduce simultaneity bias.
- Estimation strategy:
  - Dynamic specification (ARDL / ECM) to accommodate price rigidity, nonstationarity, and potential cointegration.
  - ECM estimates short-run and long-run pass-through parameters and the speed of adjustment; a negative error-correction coefficient indicates cointegration.
  - Heterogeneity across sectors addressed via Dynamic Fixed Effects (DFE) and Pooled Mean Group (PMG) estimators; Hausman test used to choose between them.

### Data
- Primary Korean data and sample:
  - Sector-level PPI and import price indices (monthly and quarterly), IO tables for benchmark years 1995, 2000, 2005, and 2010 from Bank of Korea (ECOS).
  - Quarterly sector-level ULC from Korea Productivity Center; sample restricted to 1999 onwards for thirteen 2-digit manufacturing sectors.
  - Monthly ULC series generated by cubic spline interpolation of quarterly ULC.
- Unit-root evidence:
  - Panel unit-root tests (Im-Pesaran-Shin and Fisher-type) indicate nonrejection of unit roots for sector-level price indices except for ULC (Table 1).
- Eurozone extension:
  - France, Germany, and the Netherlands using Eurostat sector-level data processed analogously.

### Key empirical findings — Korea (baseline and robustness)
- Representative aggregate statement:
  - "A 1 percent change in the effective imported input prices leads to, on average, a 0.4 percent change in domestic producer prices in the short run, and to a 0.7 percent change in the long run in Korea, with the speed of the adjustment around 1.5 percent per month."
- Baseline ECM estimates (monthly DFE, Table 2 column (1)):
  - Short-run Δln_EIMP_it (SR): 0.229 ***
  - Short-run Δln_EDOM_it (SR): 0.206 **
  - Long-run ln_EIMP_it (LR): 0.633 ***
  - Long-run ln_EDOM_it (LR): 0.789 ***
  - Error correction coefficient: -0.017 *** (implies deviations adjust by 1.7 percent per month)
- PMG baseline and variants:
  - Produce broadly similar results; Hausman tests generally fail to reject homogeneity, supporting DFE.
- Including effective unit labor cost (EULC) in the short run (Table 2 column (3)):
  - Short-run Δln_EIMP_it (SR): 0.371 ***
  - Short-run Δln_EDOM_it (SR): 0.401 ***
  - Long-run ln_EIMP_it (LR): 0.665 ***
  - Long-run ln_EDOM_it (LR): 0.863 ***
  - Error correction coefficient: -0.014 *** (implies deviations adjust by 1.4 percent per month)
- Robustness checks (Table 3, quarterly and alternative EDOM construction):
  - Quarterly DFE (Table 3 column (1)): short-run pass-through for imported and domestic inputs around 45~50 percent; long-run around 64 percent (imported) and 88 percent (domestic); error correction coefficient about three times larger than monthly estimate.
  - Monthly alternative EDOM (including own-sector input) yields short-run ~40 percent and long-run ~70~80 percent pass-through.
- Additional lags (Table 4):
  - Enrich short-run dynamics but do not overturn core results; long-run pass-through estimates remain in the ~0.63–0.88 range for imported/domestic inputs depending on specification.

### Key empirical findings — Selected Eurozone countries (France, Germany, Netherlands)
- Baseline pooled DFE results (Table 5 column (1)):
  - Short-run Δln_EIMP (SR): 0.904 ***
  - Short-run Δln_EDOM (SR): 0.563 **
  - Long-run ln_EIMP (LR): 1.057 ***
  - Long-run ln_EDOM (LR): 1.252 ***
  - Error correction coefficient: -0.045 *** (implies deviations adjust by 4.5 percent per month)
- PMG and extended-lag specifications (Tables 5–6):
  - Short-run imported input pass-through around 0.85–0.90 in extended specifications (Table 6 Δln_EIMP_it (SR) 0.854 ***).
  - Long-run ln_EIMP_it (LR) ≈ 1.017 in extended estimates.
  - Error correction coefficients around -0.044 to -0.062 across specifications (implying faster adjustment than Korea).
- Interpretation:
  - Eurozone estimates indicate substantially higher short-run and near-complete long-run pass-through of imported input cost shocks to producer prices compared with Korea, and faster speed of adjustment.

### Policy-relevant implications and conclusions
- Imported intermediate inputs are an important transmission channel for import-price and exchange-rate shocks to domestic producer prices.
- For Korea:
  - Effective imported input pass-through is incomplete: short-run roughly 0.4 and long-run roughly 0.7 in the representative statement.
  - Sector-level estimates produce short-run ~0.23–0.37 and long-run ~0.63–0.86 across specifications.
  - Adjustment toward long-run equilibrium is gradual (error-correction coefficients implying 1.4–1.7 percent monthly adjustment in preferred specifications).
- Cross-country heterogeneity:
  - France, Germany, and the Netherlands show much larger effective pass-through (short-run around 0.85–0.90; long-run ≈ 1.02–1.06) and faster adjustment (error-correction ≈ -0.044 to -0.062).
  - Heterogeneity suggests differences in market structures (e.g., competition, price flexibility) that warrant further investigation.
- Methodological implication:
  - The sector-level IO-weighted construction of effective input price indices combined with ECM estimation provides a consistent and efficient approach to quantifying imported-input pass-through and can be applied to other countries with required sector-level data.

*Source: IMF working paper content unit _wp1623 (sector-level ECM analysis of imported input cost pass-through).*

### 1. Imports of Goods by End-Use Category ..............................................................................15

### 1. Imports of Goods by End-Use Category ..............................................................................15

### Major Sections Listed
- 1. Imports of Goods by End-Use Category ..............................................................................15
- 2. Import Price Index in Korea ................................................................................................15
- 3. Imported Inputs Use in Korean Manufacturing ...................................................................16
- 4. Import Price and Producer Price Index in Korea .................................................................16
- 5. Inflation Dynamics in Korea ................................................................................................17

### Tables Included
- Table 1. Unit Root Tests for Korean Sector-level Price Series .........................................................19
- Table 2. Baseline Error Correction Model (ECM) Estimation Results for Korea .............................20
- Table 3. Extended Error Correction Model (ECM) Estimation Results for Korea (I) .......................21
- Table 4. Extended Error Correction Model (ECM) Estimation Results for Korea (II) .....................22
- Table 5. Baseline Error Correction Model (ECM) Estimation Results for Selected European  
  Countries ..............................................................................................................................23
- Table 6. Extended Error Correction Model (ECM) Estimation Results for Selected European 
  Countries ..............................................................................................................................24

*Source: _wp1623 - 1. Imports of Goods by End-Use Category ..............................................................................15*

### References .............................................................................................................

### _wp1623 - References

### Introduction and motivation
- Imported inputs are identified as the dominant transmission channel of import-price shocks to domestic prices because:
  - Consumption goods constitute at most a little over 10 percent of total imports in Korea (Figure 1).
  - Imported goods cover around 7% of Korea’s CPI basket (informal KOSTAT interview).
  - Correlation coefficient of 0.97 between monthly change of the aggregate import price index and that of the aggregate imported input price index in Korea (Figure 2).
  - Correlation of 0.63 between monthly change of the aggregate producer price index and that of the aggregate import price index; import price index is about four times more volatile than the producer price index (Figures 4–5).
- Paper focus: estimate imported input cost pass-through to domestic producer prices using sector-level monthly data and an error correction model (ECM), explicitly constructing sector-level imported input price indices using bilateral sector-level Input-Output (IO) tables and sector import prices.
- Distinct contributions:
  - Directly estimate pass-through from imported input prices to sector-level producer prices (as opposed to ERPT to import prices or CPIs).
  - Separate the roles of imported input cost shares and markup adjustments in determining effective pass-through.

### Econometric framework (baseline model and identification)
- Pricing and cost structure:
  - Firm price set as markup over marginal cost. Marginal cost modeled as a Cobb-Douglas aggregator of domestic intermediate input price (P_dom), imported intermediate input price (P_imp), and unit labor cost (ULC) with shares α, β, and γ.
  - Effective imported input cost pass-through to producer prices is (1 + ζ_β)·β where β is imported input share and ζ_β captures markup adjustment in response to imported input cost shocks.
- Empirical construction of effective input prices using IO table weights:
  - Effective imported input price EIMP_it = weighted average across input sectors of sector import price indices using imported-input shares β_ij from IO tables.
  - Effective domestic input price EDOM_it = analogous weighted average using domestic-input shares α_ij.
  - Effective labor cost term constructed as γ_i · ULC_it.
- Estimation strategy:
  - Use a dynamic specification (ARDL / ECM) to accommodate price rigidity, nonstationarity, and potential cointegration among series.
  - ECM form used to estimate short-run and long-run pass-through parameters and the speed of adjustment; when error-correction coefficient is negative it indicates cointegration.
  - Heterogeneity across sectors considered via Dynamic Fixed Effects (DFE) and Pooled Mean Group (PMG) estimators; Hausman test used to choose between them.

### Data
- Primary Korean data sources and sample:
  - Sector-level PPI and import price indices (monthly and quarterly), IO tables for benchmark years 1995, 2000, 2005, and 2010 from Bank of Korea (ECOS).
  - Quarterly sector-level ULC from Korea Productivity Center; sample restricted to 1999 onwards for thirteen 2-digit manufacturing sectors.
  - Monthly ULC series generated by cubic spline interpolation of quarterly ULC.
  - When constructing EDOM, baseline omits own-sector domestic input price to reduce simultaneity bias.
- Unit-root evidence:
  - Panel unit-root tests (Im-Pesaran-Shin and Fisher-type) indicate nonrejection of unit roots for sector-level price indices except for ULC (Table 1).
- Eurozone extension:
  - France, Germany, and the Netherlands using Eurostat sector-level data processed analogously.

### Key empirical findings — Korea (baseline and robustness)
- Main headline estimate (aggregate statement in text):
  - "A 1 percent change in the effective imported input prices leads to, on average, a 0.4 percent change in domestic producer prices in the short run, and to a 0.7 percent change in the long run in Korea, with the speed of the adjustment around 1.5 percent per month."
- Baseline ECM estimates (monthly DFE, Table 2 column (1)):
  - Short-run Δln_EIMP_it (SR): 0.229 ***
  - Short-run Δln_EDOM_it (SR): 0.206 **
  - Long-run ln_EIMP_it (LR): 0.633 ***
  - Long-run ln_EDOM_it (LR): 0.789 ***
  - Error correction coefficient: -0.017 *** (implies deviations adjust by 1.7 percent per month)
- PMG baseline (Table 2 column (2)) and variants produce broadly similar results; Hausman tests generally fail to reject homogeneity, supporting DFE.
- Including effective unit labor cost (EULC) in the short run (Table 2 column (3)):
  - Short-run Δln_EIMP_it (SR): 0.371 ***
  - Short-run Δln_EDOM_it (SR): 0.401 ***
  - Long-run ln_EIMP_it (LR): 0.665 ***
  - Long-run ln_EDOM_it (LR): 0.863 ***
  - Error correction coefficient: -0.014 *** (implies deviations adjust by 1.4 percent per month)
- Robustness checks (Table 3, quarterly and alternative EDOM construction):
  - Quarterly DFE (Table 3 column (1)): short-run pass-through for imported and domestic inputs around 45~50 percent; long-run around 64 percent (imported) and 88 percent (domestic); error correction coefficient about three times larger than monthly estimate (consistent with faster quarterly adjustment).
  - Monthly alternative EDOM (including own-sector input) yields short-run ~40 percent and long-run ~70~80 percent pass-through.
- Additional lags (Table 4) enrich short-run dynamics but do not overturn core results; long-run pass-through estimates remain in the ~0.63–0.88 range for imported/domestic inputs depending on specification.

### Key empirical findings — Selected Eurozone countries (France, Germany, Netherlands)
- Baseline pooled DFE results (Table 5 column (1)):
  - Short-run Δln_EIMP (SR): 0.904 ***
  - Short-run Δln_EDOM (SR): 0.563 **
  - Long-run ln_EIMP (LR): 1.057 ***
  - Long-run ln_EDOM (LR): 1.252 ***
  - Error correction coefficient: -0.045 *** (implies deviations adjust by 4.5 percent per month)
- PMG and extended-lag specifications (Tables 5–6) broadly confirm large pass-through:
  - Short-run imported input pass-through around 0.85–0.90 in extended specifications (Table 6 Δln_EIMP_it (SR) 0.854 ***).
  - Long-run ln_EIMP_it (LR) ≈ 1.017 in extended estimates.
  - Error correction coefficients around -0.044 to -0.062 across specifications (implying faster adjustment than Korea).
- Interpretation:
  - Eurozone estimates indicate substantially higher short-run and near-complete long-run pass-through of imported input cost shocks to producer prices compared with Korea, and faster speed of adjustment. Authors suggest this heterogeneity may reflect different market structures and price flexibility.

### Policy-relevant implications and conclusions
- Empirical evidence supports the importance of imported intermediate inputs as a transmission channel for import-price and exchange-rate shocks to domestic producer prices.
- For Korea:
  - Effective imported input pass-through is incomplete: short-run roughly 0.4 and long-run roughly 0.7 in the representative statement; sector-level estimates produce short-run ~0.23–0.37 and long-run ~0.63–0.86 across specifications.
  - Adjustment toward long-run equilibrium is gradual (error-correction coefficients implying 1.4–1.7 percent monthly adjustment in preferred specifications).
- Cross-country heterogeneity:
  - France, Germany, and the Netherlands show much larger effective pass-through (short-run around 0.85–0.90; long-run ≈ 1.02–1.06) and faster adjustment (error-correction ≈ -0.044 to -0.062).
  - Heterogeneity suggests market-structure differences (e.g., competition, price flexibility) that warrant further investigation.
- Methodological note:
  - The sector-level IO-weighted construction of effective input price indices combined with ECM estimation provides a consistent and efficient approach to quantifying imported-input pass-through and can be applied to other countries with required sector-level data.

*Source: IMF working paper content unit _wp1623 - References (sector-level ECM analysis of imported input cost pass-through).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp1623.pdf_